MOLECULAR MEDICINE: Found in Translation

Abstract

Studies of the major hemoglobin disorders, β-thalassemia and sickle cell disease (SCD), have laid a foundation for molecular medicine. While enormous progress has been made in understanding gene structure and regulation, translating molecular insights to therapy for the many individuals affected with these disorders has been challenging. Advances in three activities have recently converged to bring novel genetic and potentially curative treatments to clinical trials. First, improved lentiviral vectors for gene transfer into hematopoietic stem cells have revived somatic gene therapy for blood disorders. Second, elucidation of regulatory factors and mechanisms that control the normal developmental switch from fetal to adult hemoglobin has provided a route to reactivation of the fetal form for therapy. Third, revolutionary methods of gene engineering permit molecular insights to be leveraged for patients. Here I review how the promise of molecular medicine to bring transformative treatments to the clinical arena is finally being realized.

eTOC

Orkin discusses how recent curative genetic therapies for β-thalassemia and sickle cell disease are the result of improvements in lentiviral vectors, a detailed understanding of mechanisms that regulate the switch from fetal to adult hemoglobin, and the development of CRISPR/Cas9 gene editing.

Molecular medicine can trace its origins to studies of the inherited hemoglobin disorders. More than seventy years ago, well before elucidation of the structure of DNA or the genetic code, Pauling reported a change in composition of the β-globin chain of the adult hemoglobin tetramer (α₂β₂) in red cells of individuals with sickle cell anemia, heralding the disorder as the first “molecular disease”¹. With development of recombinant DNA methods in the late 1970s, the human globin genes were among the first mammalian genes to be cloned and then fully analyzed by DNA sequencing^2,3. At the same time, Kan and Dozy reported a restriction enzyme site polymorphism linked to a β-globin gene harboring the sickle cell mutation, representing the first clinically relevant disease association made through DNA analysis⁴, and a primitive harbinger of contemporary genome-wide association studies (GWAS). Comprehensive discovery of the diverse mutations in β-thalassemias, inherited disorders of β-chain production, in the 1980s provided the first nearly complete description of a disease at the molecular level, and a critical database for effective DNA-based prenatal diagnosis⁵.

Research focused on globin genes and developing red blood cells has informed concepts of gene structure and expression, chromatin dynamics, and the role of transcription factors in lineage determination and differentiation. The major cis-regulatory element for expression of the β-globin genes, an extended enhancer known as the locus control region (LCR)^6,7, was the first experimentally established super-enhancer⁸. The looping of the LCR to downstream globin genes has formed the basis for current views of gene activation by enhancers ^9,10 and the higher order structure of chromatin and its interactions. Study of the master erythroid transcription factor GATA1 revealed how red cell gene expression is programmed ^11,12, and led to the identification of rare inherited red cell and platelet disorders, as well as an underlying somatic mutation characteristic of preleukemia and myeloid leukemia in the setting of trisomy 21^13,14.

While these discoveries illustrate the extraordinary power of contemporary molecular methods to dissect genetic diseases and underlying mechanisms, patients rightfully ask “what have you done for us?” Research grant applications promise translation of genetic discoveries to patients, but the sobering reality is that the route to the clinic and positive clinical benefit is challenging and often not straightforward. The early history of genetic therapy for hemoglobin disorders mirrors the development of somatic gene therapy for other inherited conditions: slower than originally anticipated and marred by egregious investigator misjudgments. Soon after the cloning of globin gene sequences and prior to any understanding of how the genes are regulated, globin cDNAs in a plasmid with a selectable marker were transfected into bone marrow cells, which were then infused into patients in a human experiment that became known as the “Cline affair”¹⁵. Fortunately, in the intervening years other investigators pursued evidence-based approaches in a decades long path to rational and effective therapy.

The Hemoglobin Disorders

Although β-thalassemia and sickle cell disease (SCD) are granted “orphan” disease status for the purpose of encouraging development of new treatments, they are by no means rare disorders. Together, hundreds of thousands, or millions, of individuals worldwide are affected with two β-thalassemia or sickle (β^S) alleles. In sub-Saharan Africa, ~300,000 or more children are born with SCD each year and sadly face >75% mortality by age five largely due to infectious disease. In the United States, the prevalence of SCD is 75–100,000 individuals with an annual health expenditure of ~$1B. The global disease burden of hemoglobin disorders is therefore huge, and anticipated to increase in the coming years^16,17.

The β-thalassemias, which result from deficiency (β⁺) or complete failure (β⁰) in production of the β-globin chain of adult hemoglobin (α₂β₂), are due to diverse mutations in the β-globin gene affecting its transcription, RNA processing, or translation of its mRNA into protein^5,18. Although SCD is due to a single amino chain in the β-chain (G>V at position 6), which renders sickle hemoglobin (α₂β^S₂) prone to polymerize at low oxygen, the spectrum of clinical severity is notoriously broad. Management of patients with β-thalassemia and SCD has traditionally been largely supportive. For β-thalassemia, red cell transfusion and iron chelation are mainstays of therapy. In SCD, management consists of pain control and penicillin prophylaxis for functional asplenia, supplemented by chronic transfusions for those with intractable pain, stroke, or splenic sequestration. As general medical care in resource-poor countries improves, a marked increase in disease prevalence will further tax health systems. Standard of care for SCD includes oral treatment with hydroxyurea (HU), the only agent with proven substantial clinical benefit^19,20. Replacement of the hematopoietic system by marrow transplantation is curative in both β- thalassemia and SCD, and limited by availability of histocompatible donors, the inherent toxicity of myeloablative preconditioning of the recipient, and the overall resource intensive nature of the procedure.

Addressing the primary problem in the hemoglobinopathies

The pathophysiology of the disorders is well understood^18,21. In β-thalassemia, deficient or absent β-globin production establishes an imbalance between α-globin and β-globin chains, leading to precipitation of a-globin, hemolysis, and ineffective erythropoiesis. In SCD, polymerization of HbS impairs developing red cells, leading to loss of red cell deformability, microvascular damage, anemia, and the secondary complications of pain and stroke. While therapies may be directed to downstream effects of chain imbalance and sickling, it is axiomatic that targeting the hemoglobin composition should be the most direct and effective means of lessening disease severity and forestalling complications. Well before the contemporary molecular era, astute clinical investigators provided clues to possible solutions. The wide range of severity of the hemoglobinopathies points to the contribution of genetic modifiers. The two strongest modifiers are α-thalassemia (decreased α-globin production) and the level of fetal hemoglobin (HbF). In β-thalassemia, α-thalassemia reduces globin chain imbalance and subsequent hemolysis and ineffective erythropoiesis, despite lowering of the total hemoglobin concentration in the red cell. Since sickling of HbS is concentration-dependent, α-thalassemia is also beneficial in SCD²². Increased HbF replaces missing HbA in β-thalassemia, and in SCD elevated HbF has a two-fold benefit. First, HbF blocks HbS polymerization and second, β^S- globin production is reduced concomitantly with increased γ-globin expression due to the reciprocal nature in which the linked genes are transcribed in the human β-like globin complex. The salutary effect of HbF on SCD was first deduced by the pediatrician Janet Watson in 1949²³. She observed that children with SCD were asymptomatic in the first few months of life and hypothesized that HbF was protective.

Progress in genetic approaches to the hemoglobin disorders that are likely curative has accelerated greatly in recent years, largely due to development of improved vectors for gene transfer into hematopoietic stem cells (HSCs), the advent of transformative gene editing technologies^24,25, and a deeper understanding of globin gene regulation.

Globin gene regulation and the fetal-to-adult switch

The organization and expression of human globin genes has been a topic of intense investigation since the molecular cloning of the α- and β-globin gene clusters in bacteriophages in 1978^2,3. This landmark achievement in the laboratory of Tom Maniatis was reported at the first Hemoglobin Switching Conference, a biannual meeting organized by Arthur Nienhuis and the late George Stamatoyannopoulos, and supported by the NIH, that has fostered development of the field to the present²⁶. The name given to the conference was prescient as it embodied a conviction that understanding how globin genes are regulated would have clinical impact. As the major hemoglobin disorders are due to β-globin gene mutations, only the human β-globin complex on the short arm of chromosome 11 will be discussed further in this Perspective.

Within the β-globin cluster, the β-like genes are arranged downstream of the LCR in the order in which they are expressed during development: LCR- embryonic (ε ) – the duplicated fetal (Gγ–Aγ) –adult (β) —genes (Figure 1) Activation of transcription of each gene is achieved through physical looping and contact with the LCR. The first globin switch, embryonic ε– to fetal γ-globin occurs in the transition to fetal liver hematopoiesis in the second trimester. Thereafter, the critical fetal-adult switch takes place gradually and is completed only in the first postnatal months. In the adult, the level of HbF is ~1%, which is largely restricted to a small population termed F-cells²⁷.

Figure 1. The fetal-to-adult hemoglobin switch.

The critical transition from HbF(α₂γ₂) to HbA (α₂β₂) entails a transcriptional switch from engagement of the locus control region (LCR) with the γ-globin genes to the β-globin gene. Two repressor proteins BCL11A and LCR (ZBTB7A) initiate the switch by blocking γ-globin transcription and preventing access by the LCR. Repression of γ-globin gene expression allows interaction of β-globin gene with the LCR and transcription activation.

In adults, the level of HbF is under genetic control of two kinds. Common genetic variation leads to quite modest changes (few %) in HbF. Rare genetic variation, deletions of DNA within the cluster or specific single nucleotide changes in the γ-globin promoters, is associated with substantial (10–30x normal) HbF production, termed hereditary persistence of fetal hemoglobin (HPFH)²⁸(Figure 2) The point mutations in the γ-promoter exert profound effects, such that the level of expression of the linked γ-globin gene is expressed at full level. Therefore, inheritance of an HPFH allele in trans to either a β^S or β-thalassemia allele greatly counteracts either chain imbalance or sickling, and may eliminate disease manifestations.

Figure 2. Mechanism of γ-globin gene repression.

The sequence of the γ-globin gene promoter is depicted. The DNA-binding motifs of the two repressors LRF/ZBTB7A and BCL11A are highlighted in blue and red, respectively, with base substitutions observed in patients with HPFH noted below the sequence. Both BCL11A and LRF physically interact with the multicomponent NuRD complex. Binding of BCL11A and NF-Y, an activator that binds a CCAAT box (underlined in yellow) that overlaps a proximal BCL11A motif, oppose each other’s action. BCL11A displaces NF-Y by steric hindrance in the promoter. The boxed region above the sequence indicates a 13bp deletion observed in HPFH and also generated frequently upon CRISPR/Cas9 editing of the promoter ⁶⁸. Figure modified from Orkin and Bauer ⁵⁵.

Expression of lineage restricted genes in developing red cell precursors is under the control of the major erythroid transcription factors (TFs), including the master regulator GATA1 and additional factors (TAL1, NF-E2, KLF1)²⁹. None of these TFs exhibit stage-selective expression or activity that can account for the silencing of γ-globin expression in the fetal-adult switch. The search for candidate TFs controlling the switch proved elusive.

Application of GWAS to identify candidate loci regulating HbF level shined new light on the problem. Cotemporaneous studies by Menzel and Thein²⁹ and by Uda and associates³⁰ revealed that a previously unsuspected locus on the short arm of chromosome 2, BCL11A, was strongly associated with F-cell number and HbF level, respectively. SNPs near or within the BCL11A gene reflect common genetic variation that is widely distributed among populations, i.e. they are not restricted to those groups in which β-thalassemia or SCD are prevalent. In concert with other SNPs in the vicinity of the c-Myb locus and within the β-globin gene cluster, such common variation may explain as much as 25–50% of genetic variation of HbF^30,31, which is enormous in relation to the majority of genetic contributions attributed to GWAS in most disease settings.

BCL11A has served as an entry point into the mechanistic details of the hemoglobin switch. Consistent with a postulated role as a repressor of γ-globin expression, BCL11A is not expressed as RNA or protein at the embryonic stage of erythroid development, and first appears at the time of fetal liver erythropoiesis in the definitive (or adult) lineage of erythroid cells³². Within human fetal liver erythroid cells, BCL11A mRNA translation is partially blocked by RNA- binding proteins that are subsequently down-regulated at the adult stage³³. Down-regulation of BCL11A by short-hairpin RNA (shRNA) in adult erythroid cells (derived from CD34+ stem/progenitor cells) leads to increased HbF, compatible with the proposed action of BCL11A as a γ-globin gene repressor³². In retrospect, given the myriad roles that many TFs play in erythroid cell differentiation and the effects of many parameters, including cell proliferation and maturation, on the relative level of HbF in such cultured cells, these early experiments might only be accepted as successive evidence in favor of BCL11A as a potential therapeutic target.

A priori, silencing of γ-globin gene transcription might rely on a few, or numerous, stage- selective factors. The number of factors involved has direct implications for any target-based reactivation of HbF. If a single factor (or just a couple) is critical, a target-based strategy might be feasible. On the other hand, if many factors are involved, prospects for such therapeutic approaches would be remote. Fortuitously, the former situation appears to apply to control of HbF, as manipulation of BCL11A alone is sufficient to achieve quite high level derepression. The rescue of hematologic parameters of engineered sickle cell mice by erythroid-specific knockout of BCL11A provided crucial insight³⁴. Engineered sickle cell mice lack mouse globin genes and express human globin genes, including the β^S-gene, from transgenes^35,36. Knockout of BCL11A in the erythroid lineage led to nearly complete rescue of red cell parameters, and pancellular reactivation of HbF to ~30% of total Hb. Thus, removal of the single factor was sufficient to achieve phenotypic rescue. Given that the human globin genes in engineered mice mimic but do not fully recapitulate the normal organization of globin gene clusters, it is not possible to predict whether ~30% HbF constitutes an upper limit for reactivation in a human erythroid cell lacking BCL11A or a lower limit. Nonetheless, the engineered sickle cell mouse experiments provided a persuasive rationale for pursuit of BCL11A as a genetic target for the hemoglobin disorders.

HbF levels are quite sensitive to the expression of BCL11A (Figure 3). Common genetic variation detected in GWAS is associated with a modest (~40% estimated) reduction in BCL11A expression and a small increase in HbF³⁷. Moreover, in patients with the autism-like BCL11A haploinsufficiency syndrome HbF levels average ~10–15% with a wide range^38–40. Highly associated SNPs lie within the second intron of the BCL11A gene, within and surrounding, an ~10 kb erythroid-specific enhancer element characterized by 3 subregions (+55, +58, +62) marked by DNase I hypersensitivity³⁷. A causal SNP in the +63 HS disrupts a canonical GATA-TAL1 binding site. Despite the large size of the BCL11A erythroid enhancer, a GATA1 binding site within +58 confers a major portion of activity to the entire enhancer⁴¹. This “Achilles heel” represents an especially favorable target for genetic disruption by either zinc-finger nucleases or CRISPR/Cas9^41–43. Besides the marked reduction in BCL11A expression observed upon targeting these sequences, the effects of the edits on expression are entirely restricted to erythroid cells⁴⁴. Cell-selectivity is important as BCL11A is expressed and critical for proper hematopoietic stem cell function and B-lymphoid development, though the relative thresholds for impairment in these cells are unknown^45,46. For example, in haploinsufficient individuals, B-cell development and overall hematopoiesis appear normal³⁸.

Figure 3. Genetics of BCL11A and HbF output.

The consequences of genetic alterations at the BCL11A locus for HbF expression are summarized. The panels depicting common low and high HbF genotypes at the BCL11A locus illustrate the effect of common genetic variation, as measured in GWAS, on BCL11A expression and HbF expression. Natural variants at BCL11A lie within the erythroid-specific enhancer of the BCL11A gene and determine the level of transcription and output of BCL11A protein. The “high HbF” allele is associated with reduced BCL11A protein and a modest increase in HbF. Deletion of the erythroid-specific enhancer nearly abolishes BCL11A expression, leading to marked increase in HbF, as shown in the third panel. Therapeutic gene editing of the erythroid enhancer, specifically at the “Achilles heel” region, greatly reduces BCL11A expression but not to the extent of the enhancer deletion. Nonetheless, the level of HbF expression is significantly increased and is sufficient to ameliorate the severity of either β-thalassemia or SCD. Figure modified from Hardison and Blobel ⁸⁷.

Although potent and essential as a regulator of Hb silencing, BCL11A is not the sole factor. A second factor, named ZBTB7A or LRF, is comparable in its activity, as shown by knockout in immortalized CD34+ derived erythroid cells (HUDEP-2)⁴⁷. Knockout of either BCL11A or LRF leads to ~40–50% HbF in this model. HbF levels are 90–100% in cells with combined knockout. Both factors physically associate with the NuRD corepressor complex, which confers both remodeling and histone deacetylase activities^47,48. CRISPR/Cas9 mutagenesis of the various subunits of NuRD demonstrates that specific paralogs of each subunit are required for HbF repression⁴⁹. Further support for a cofactor role for the NuRD complex in HbF repression is provided by recent findings that ZNF410, a pentadactyl DNA binding protein, exhibits remarkable specificity for activation of the CHD4 locus, which encodes its ATP-dependent remodeling subunit^50,51.

Recent studies provide an increasingly detailed view of how γ-globin gene repression is controlled (Figure 2). Whereas chromosomal looping between the LCR and each downstream globin gene is believed to be required for transcriptional activation, the choice between expression of γ- and β-globin genes appears to be executed locally. Specifically, γ-globin gene repression depends on the two repressors (BCL11A and LRF), each binding to a cognate recognition site within the γ-globin promoter^52,53. Three closely spaced zinc-fingers at the C-terminus of BCL11A specific recognition of a unit motif, TGACCA, which is duplicated in γ-globin promoter. Remarkably, BCL11A binds selectively to the more distal of the duplicated sites, and some of the rare, naturally occurring point mutations leading to the HPFH syndrome disrupt the TGACCA motif⁵³. The duplicated TGACCA motifs overlap duplicated CCAAT boxes, recognition sequences for the ubiquitous CCAAT-binding protein complex NF-Y. Of note, in cells expressing γ-globin, NF-Y selectively binds the more proximal CCAAT motif⁵⁴. Detailed functional studies indicate that BCL11A binding to its cognate TGACCA site displaces NF-Y from its proximal site to initiate repression. Recruitment of the NuRD complex by BCL11A is believed to stabilize repression. LRF recognizes a GC-rich motif at a more distal position in the γ-globin promoter, which is also the site of rare single base changes in HPFH⁵². As LRF also associates with the NuRD protein complex, it is likely that concerted recruitment of NuRD to the promoter is required for stable repression⁵⁴. Such a concerted action may explain how loss of either BCL11A or LRF leads to a similar deficit in repression, whereas combined loss further augments repression.

Strategies for genetic management of hemoglobin disorders

Disease pathophysiology can be reversed only by addressing the quality and quantity of hemoglobin within red cells. In β-thalassemias, raising the level of β-like globin (either β- or γ-globin) is the goal. As the relative imbalance of α- and β-globin chains is reduced, the extent of ineffective erythropoiesis and anemia is lessened. Approaches to genetic therapy include correcting the specific mutation for the given thalassemia allele(s), expression of an exogenously introduced β-like globin gene, or reactivation of endogenous γ-globin to produce high levels of HbF⁵⁵. In principle, correction of the underlying sickle mutation is the simplest approach as the same base change is present in all β^S alleles. Expression of a non-sickling β-globin or γ-globin, or reactivation of endogenous γ-globin, is more tractable with current technology.

Transcriptional output from the human β-globin cluster is directed by the LCR, which physically loops to downstream globin genes for activation. As such, expression from the γ- and β-globin genes is reciprocal. In fetal erythroid cells, the LCR loops to the γ-globin genes, and β-globin transcription is largely prevented. Binding of BCL11A and LRF to the γ-globin promoters down-regulates transcription, and frees up the LCR (by unknown mechanisms) to loop to the β-globin gene. Reactivation of γ-globin expression upon removal of BCL11A (or LRF) leads to concomitant down-regulation of β-globin expression. As a practical consequence of these mechanisms, reactivation of the endogenous γ-globin genes in SCD reduces HbS production. In effect, increased HbF affords two benefits in SCD: first, HbF impairs the sickling process and second, HbS production itself is turned down. Similarly, in β-thalassemia, increased expression of endogenous HbF replaces missing HbA (from β-globin deficiency) and down-regulates β-globin production which lessens chain imbalance and ineffective erythropoiesis.

The following strategies are available for genetic therapy of the hemoglobin disorders (Table 1):

1. Globin gene addition: introduction of a vector that provides for expression of an exogenous globin (either β-globin, γ-globin, or a non-sickling β-globin)

2. Gene correction: introduction of gene modifying reagents to correct the underlying gene mutation

3. Reactivation of endogenous HbF: introduction of vector or gene modifying reagents that lead to re-expression of γ-globin genes in adult erythroid cells due to either down-regulation of a repressor (BCL11A) or alteration of a repressor binding site (i.e. recreation of an HPFH allele).

Table 1. Targets for gene manipulation therapy.

Proposed targets for genetic treatment of β-thalassemia and SCD are presented with the corresponding method, approach, and desired outcome. HDR: homology directed repair; NHEJ: non-homologous end-joining.

Target	Method	Approach	Outcome	HbF expression	Clinical trial
β-globin gene	CRISPR/Cas9 (or similar)	HDR	Corrected gene	No change
BCL11A mRNA	Lentivirus	shRNA-mediated erythroid knockdown	Reduced BCL11A expression	Reactivation	Yes
BCL11A erythroid enhancer	CRISPR/Cas9 (or other editing)	NHEJ	Reduced BCL11A expression	Reactivation	Yes
BCL11A binding site in γ-globin gene promoters	CRISPR/Cas9 (or other editing)	NHEJ	Mimic HPFH allele	Reactivation
BCL11A binding site in γ-globin gene promoters	Base editing	C>T conversion	Mimic HPFH allele	Reactivation

Genetic technologies

At present, genetic therapy for hematopoietic disorders requires ex vivo modification of enriched stem/progenitor cells (CD34+ cells) and preconditioning of the host with chemotherapy to ablate endogenous hematopoiesis. For a durable result, gene transfer into long-term repopulating hematopoietic stem cells (HSCs) is critical. Early studies in gene therapy employed retroviral vectors. Inefficient transduction of HSCs, shutoff or inconsistent vector expression, and insertional mutagenesis plagued early clinical trials⁵⁶. Transition to use of lentiviral vectors, which exhibit more random insertion into the genome and stable expression, has greatly accelerated progress in the field.

Gene modifying strategies have undergone a revolution in recent years^24,57. Zinc-fingers and TALENs allowed “proof-of-principle” experiments for gene disruption by non-homologous end joining (NHEJ) and by homology-directed repair (HDR). Indeed, knockout of the HIV CCR5 receptor in lymphoid cells with zinc-fingers constituted the first human gene editing trial⁵⁸. The advent of CRISPR/Cas9 editing has brought gene editing to the fore, given its ease of use, flexibility, high efficiency and low off-target action. The newer method of base editing expands options for gene manipulation. As discussed below, critical issues in clinical application include efficiency, off target effects, and the nature of the target under consideration for therapeutic intervention.

Clinical trials for hemoglobinopathies (Table 2)

Table 2. Gene therapy approaches in clinical development for hemoglobinopathies.

Data are derived from recent publications and ClinicalTrials.gov as of November 2020.

Molecular approach	Disorder	Clinicaltrials.gov identifier	Sponsoring institution	Approval
Globin addition (modified β) Lentivirus	SCD β-thalassemia	NCT02140554 NCT03207009 NCT02906202	Bluebird bio	ZYNTEGLO® (Europe)
Globin addition (modified β) Lentivirus	SCD	NCT02247843	UCLA
Globin addition (modified γ) Lentivirus	SCD	NCT02186418	Cincinnati Children’s Hospital Medical Center
HbF reactivation Erythroid shRNA knockdown of BCL11A Lentivirus	SCD	NCT03282656	Boston Children’s Hospital/Bluebird bio
HbF reactivation Gene editing of BCL11A enhancer CRISPR/Cas9	β-thalassemia SCD	NCT03655678 (Climb THAL-111) NCT03745287 (Climb SCD-121)	CRISPR Therapeutics/Vertex

Globin gene addition:

The first promising clinical trials for hemoglobinopathies have relied on lentiviral vectors in which a β-globin gene is expressed under the control of elements derived from the β-globin cluster LCR. After a long period of vector development, lentiviruses that efficiently transduce HSPCs and express β-globin were identified. In 2010 LeBoulch and colleagues reported a single patient with transfusion-dependent β^E/β⁰-thalassemia, the most common form of β-thalassemia in Southeast Asia, that was treated with a lentivirus expressing a mutated β-globin β^T87Q with anti-sickling properties⁵⁹. Following allogeneic hematopoietic reconstitution with modified HSPCs, the patient became transfusion-independent and maintained a Hb of 9–10 g/dl, of which ~30% was derived from the vector. The majority of exogenously expressed globin was derived from a dominant, myeloid-based clone, which was expanded due to insertional HMGA2 gene activation in erythroid cells. Clonal dominance was lost by 12 years post therapy and associated with the need for intermittent transfusions⁶⁰.

The largest clinical experience in globin addition therapy is that reported by Cavazzana and colleagues in which 22 patients with transfusion-dependent β-thalassemia were treated with LentiGlobin BB305, a vector that was slightly modified from the used in the earlier LeBoulch patient⁶⁰. Patients with β⁺-thalassemia (i.e. those with at least one non-β⁰-allele) fared significantly better than β⁰-thalassemia individuals, of which only 3/9 became transfusion-independent. Clinical benefit was documented as transfusion requirements were reduced by 73% overall with associated improvement in hematologic parameters. No serious adverse events were reported. Despite the positive findings of this trial, full correction of pathophysiology was not achieved. LentiGlobin BB305 has been approved in Europe as treatment for β⁺-thalassemia patients >12 year of age. In a separate trial (TIGET, BTHAL, NCT02453477), which is no longer recruiting patients, 3 adults and 6 children with β-thalassemia were treated with a β-globin lentivirus (GLOCE) by intrabone administration of transduced HSCs⁶¹. Transfusion requirements were reduced but not eliminated in the adults and 3 of 4 evaluable pediatric patients became transfusion independent.

LentiGlobin BB305 has also been used in a similarly designed trial to treat patients with SCD. A single patient has been described with correction of hallmarks of the disease⁶². A recent update of the trial, now including a total of 14 SCD patients, described resolution of sickle crises in almost all patients⁶³.

BCL11A-directed reactivation of HbF:

The alternative strategy of increasing endogenous HbF to replace missing HbA in β-thalassemia or retard sickling in SCD has the added theoretical benefit of leveraging the reciprocal pattern of expression of the γ- and β-globin genes within the globin cluster to reduce transcription of the mutant β-gene. The most critical parameter influencing the potential success of this approach is the level to which HbF can be raised with an intervention. Based on preclinical studies, near complete loss of BCL11A in erythroid precursors would likely reactivate HbF to ~40–50% of total Hb in a given modified cell, a level well within a therapeutic range for either β-thalassemia or SCD. Recreation of a single HPFH mutation of the “Greek”-HPFH variety would be similarly beneficial.

Lentiviral shRNA to BCL11A:

Based on preclinical studies showing that either erythroid-specific knockout of BCL11A or erythroid-directed lentiviral shRNA directed to BCL11A reactivates HbF to levels sufficient to rescue SCD mice^46,64, a clinical trial was designed and initiated at Boston Children’s Hospital. The trial is innovative in several respects. It is the first trial in which an shRNA, in this instance was embedded in a miRNA scaffold (a shmiR), has been employed. Second, the shmiR was contained within a lentiviral vector similar to that used in the globin addition trials in order to limit expression to erythroid precursors. Third, the trial represents the first in man to assess reactivation of endogenous HbF. The initial results of the trial, which has recently been reported⁶⁵, are remarkable for the level of HbF reactivation observed and for the overall consistency patient to patient. Of 6 patients 6–21 months post therapy, F-cell level were ~70% with F/F-cell = 9.6–13.8 pg (33–47% of total Hb/cell) with a total HbF of 22–40%. All patients became symptom-free and no serious adverse events were observed. Hematological parameters indicated some residual ongoing hemolysis but very much reduced from pre-therapy values. HbF was not detected in ~25% of red cells, i.e. that they were the product either of unmodified HSCs or from residual host HSCs that were not ablated in preconditioning. The former possibility is more likely but additional studies are required. The initial findings of this trial are important in that they fully validate the preclinical science underlying the overall strategy.

Gene editing:

Preliminary findings of a collaborative trial between CRISRP Therapeutics and Vertex to reactivate HbF by CRISPR/Cas9-mediated disruption of the “Achilles heel” in the BCL11A enhancer⁴¹ have been reported recently^66,67. In this instance, Cas9 protein and an sgRNA in an RNP complex was electroporated into HSPCs ex vivo and then reinfused into preconditioned 5 β-thalassemia or 2 SCD patients in the trials to date. The results revealed thus far are very promising. All patients with transfusion-dependent β-thalassemia remained transfusion-free from 2 months post-therapy. The 2 patients with SCD patient experienced no vaso-occlusive crises post-therapy. All treated patients maintained near normal hemoglobin levels and in the two individuals with SCD HbF was >40%. To date, no serious adverse events have been reported. These first-in-human CRISPR/Cas9 editing trials of HSPCs appear to demonstrate clinical benefit of gene editing in the hemoglobin disorders. Additional clinical trials based on gene editing of the BCL11A enhancer are anticipated from other academic and commercial entities.

In development:

At least three other approaches are in preclinical development. First, based on the critical role of the BCL11A binding site in the γ-globin promoters, CRISPR/Cas9 gene editing to disrupt the motif will likely be brought forward as a strategy to recreate an HPFH-like allele in patients. The presence of duplicated sequences in the promoters tends to favor a 13 bp deletion within the promoter or a larger deletion that reduces an allele to a single γ-globin gene ⁶⁸, although specific gRNAs or choice a different Cas9-like protein permits discrete editing of the motif. Second, the BCL11A binding motif is also favorable for use of base editors that convert C-T^24,69. In this setting the site can be mutated to prevent occupancy of BCL11A protein, and therefore mimic the HPFH syndrome. Base editing has a theoretical advantage over gene editing in that base conversion does not involve a cut in the target DNA and therefore may avoid cellular responses to DNA breaks and therefore afford greater safety. Base editing to create other HPFH-like mutations in the γ-globin promoter can also be envisioned. Third, precise correction of the sickle mutation or any given β-thalassemia mutation by HDR is in theory the most direct approach as it converts the target allele to the normal β-globin sequence. To date, attention has focused largely on correction of the sickle mutation using CRISPR/Cas9 and related procedures^70,71. The challenges to clinical application of HDR for SCD relate to the overall efficiency at which the correction is achieved and the concomitant generation of indels, which in effect convert the β^S gene to a β-thalassemia allele. Improvements in methodology have greatly improved the frequency of gene correction, and therefore it is likely that this approach will be brought to clinical trials in the future⁷².

Lessons from clinical trials

Several lessons are emerging at this early stage of clinical translation:

First, current technology --either lentiviral gene therapy or gene editing-- is capable of achieving substantial improvement in hematological parameters and clinical benefit in patients with β-thalassemia and SCD. To be sure, further studies are required to assess durability, consistency of responses, and safety of current approaches. Nonetheless, the demonstration that transfusion requirements can be alleviated in β-thalassemia and sickle crises reduced or eliminated in SCD constitutes a significant milestone in genetic therapy. With a goal of complete restoration of normal red cell production and dynamics, there is room for improvement.

Second, achieving meaningful therapeutic benefit necessitates modifying a substantial proportion of the host HSCs. Based on the findings in SCD patients who have exhibited mixed chimerism after bone marrow transplantation from a compatible donor, discussion in the field has centered on what might constitute the lowest threshold for the proportion of gene modified cells for clinical benefit^73,74. A threshold of 10–20% has generally been taken as a consensus estimate that a given manipulation of HSPCs ex vivo – gene transduction, gene editing, gene correction by HDR -- might likely be in “therapeutic range”. Preclinical and clinical data thus far argue however that higher levels will be needed to achieve transformative and consistent benefit to patients. In lentiviral globin addition trials, enhanced HSC transduction has been critical in both β-thalassemia and SCD. In preclinical experiments, gene editing frequencies of 70% or greater are required to obtain consistent and robust results in reconstituted mice⁴².

Third, reactivation of endogenous HbF as a therapeutic strategy for β-thalassemia and SCD, a long-sought goal envisioned decades ago, is feasible and likely to provide results comparable to that of globin gene addition. The in vivo findings in patients are remarkably consistent with preclinical studies involving manipulation of BCL11A^34,64,65. Whether the theoretical benefit of greater physiologic control of globin gene expression achieved by reactivation of HbF (as opposed to globin addition) is realized in clinical outcomes remains to be explored. Validation of BCL11A as a therapeutic target in patients also supports a rationale for small molecule-based therapeutics directed to the protein as an alternative, and more scalable, strategy to reduce disease burden in populations.

Challenges going forward

In the short term, the durability and safety of the ongoing gene-based trials need to be assessed carefully. Given the different therapeutic strategies, comparison, one-to-another and head-to-head where possible, will be important in assessing the best options for patients. The relative benefits and risks of the different approaches merit careful consideration. Although the current generation of lentiviral vectors has an increasingly good risk profile, the specter of insertional mutagenesis or oncogene activation can never be entirely eliminated^75,76. In principle, transcriptional repression of the inserted lentivirus long after therapy is possible. In trails employing CRISPR-based editing, off-targeting by sgRNAs and larger scale chromosomal alterations and losses must be assessed in the setting of clinical trials^77,78. In a true sense, as we enter the realm of therapeutic genome modifications we participate in a game of roulette where the risk of an intervention must be weighed against the morbidity of the target disease in the background of ongoing mutation of the genome that occurs with natural aging⁷⁹.

As therapies move to approval, the ease and cost of preparation of molecular reagents versus large scale lentivirus production may impact commercial competitiveness. Apart from these important aspects, ex vivo genetic therapy for hemoglobinopathies is auto-transplantation with modified HSPCs, and therefore resource intensive. Innovative ways of reimbursement are needed in order to reduce inequities in availability of novel therapies to patients and yet encourage commercial investment in development of these treatments for “orphan” diseases⁸⁰.

Chemotherapeutic preconditioning currently employed for myeloablative transplantation, as used in gene therapy, carries potential risks for patients. Non-ablative preconditioning with antibodies or antibody toxin conjugates may offer safer and less toxic alternatives in the future^81,82. Instead of HSPCs harvested from patients, HSCs generated from pluripotent stem cells (iPSCs) that have been genetically corrected in culture could in principle constitute an alternative source of cells for hematopoietic reconstitution⁸³. For this goal to be realized, however, major enhancements in derivation of authentic HSCs and their expansion are needed⁸⁴.

As currently executed, genetic therapy for β-thalassemia and SCD cannot adequately address the true burden of disease either in the US or elsewhere, and particularly in resource-limited areas¹⁷. In vivo genetic therapy, in which a viral product or a gene modifying complex is delivered once or repeatedly, is envisioned as a possible approach^85,86. Rapid progress has been made in in vivo genetic therapy for eye disorders and prospects for treatment of disorders for which correction of cells in the liver is beneficial are exciting. Given the need to modify HSCs at a substantial level to achieve lessening of disease pathophysiology in thalassemia and SCD, transformative technologies will be needed to realize this vision for blood disorders.

Conclusion

While much remains to be done, advancing genetic therapies to patients with β-thalassemia and SCD once again affirms the inestimable value of long-term investment in fundamental science. Besides the benefit afforded to patients with these disorders, experience in the various approaches to gene transfer and gene modification will accelerate progress toward innovative therapies for other conditions that may be less common and attract fewer resources. The field of molecular medicine is at an inflection point in the development of curative genetic therapies.